Cage-Yoda is more Cage than Yoda, According to Google’s Vision API
Google’s Vision API is good. It can identify faces, landmarks, animals, vegetables, emotions, and more. It is so advanced that I decided to put it up against the ultimate test.
The Nicholas Cage Yoda test. Behold.
Is it Nicholas Cage? Is it Yoda? It’s hard to know. After passing the image through Google’s Vision API, here is what it thinks.
Applause to Google! They identified not only the outline of the ancient Jedi Master, but were also able to ferret out Nicolas Cage’s face.
To be clear, the output above is from the Web detection part of the Vision API. Google says that ‘Web Detection detects web references to an image’, and so it’s probably looking at the content on the websites that it finds this image is related to. Since the this Yoda and Cage mashup is on a lot of web pages that talk about both Nicholas Cage and Yoda, perhaps that’s how it made the inference.
Still, it helps us properly name the creature in the picture. According to the score, it’s slightly more Nicholas Cage than it is Yoda. We therefore have a Cage-Yoda, and not a Yoda-Cage.
What’s more, Google gives us a hint at his emotions.
It is ‘very likely’ that Cage-Yoda, who appears to be preparing for light saber combat, is feeling joy. Probably because Cage-Yoda always wins, and knows this immutable fact about himself.
But just how recognizable is Nicolase Cage under different circumstances? What about Rabbit-Cage?
More like Cage-Rabbit. In fact, this one is way more Cage than it is Rabbit. Look at the scores – Nicolas Cage at 14.9, Rabbit at 0.7. He’s 21 times more Cage than rabbit. Maybe the ferocity of emotion being displayed imbues the image with so much Cage there isn’t room for much else.
In fact, Cage’s intensity is so overwhelming that the labeling function of the Vision API also gets confused.
No sign of the rabbit! It gone blown away. This is either a Cage-Macaque or a Cage-Old World Monkey. Not being sure what either of those mean, I think I’ll just stick with calling this one the Cage-Mammal.
So Nicolas on a Jedi and Nicholas on a mammal are both detected. But what about Nicolas on my brother-in-law?
Sloppy photo editing skills aside, at least we have a picture Google won’t recognize anywhere on the web for help since I took this one myself. For this picture, the algorithm is on its own.
Wow! Cage is nowhere to be found. But Robert Covington is. Apparently the equation is:
My brother-in-law + Nicolas Cage’s face = Boss Basketball Player.
Which now begs the question, who is contributing more to the baller skills? Is it my brother-in-law, or is it the Cage? Like I learned back in the chemistry lab, the only way to tell is by isolating the two entities.
Here is isolated Cage.
Pretty much all Cage, with a hint of sorcery and a pinch of actor. Sounds about right.
Here is isolated brother-in-law.
There it is. Ben Simmons. All the baller power seems to emanate from brother-in-law. It’s a little bit unclear what it is about adding Cage to the mix that makes Ben Simmons turn into Robert Covington, but one theory could be Cage’s additional facial hair growth.
Whatever the case may be, don’t mess with my brother-in-law. Not only does he have mad game, but according to Google, he’s also swol (i.e. muscular, for our less street-savvy readers).
These few Nicholas Cage experiments are just the tip of the iceberg. Check out the Vision API for yourself – you can test out images for free. If you find anything interesting, let us know in the comments! And if anyone from Google is reading this, I’d love your perspective on the above interpretations of the API’s output.